This proposal uses a randomized trial design to test three computer- tailored self-help interventions designed to increase rates of cessation among cigarette smokers calling the Cancer Information Service (CIS) for help in quitting. CIS callers will be randomly assigned to one of four experimental conditions: 1) Brief Educational Message (BEM) promoting smoking cessation delivered by CIS Information Specialists over the telephone, combined with a single non-tailored follow-up mailing of reinforcing print material (this is the control condition); 2) BEM delivered by CIS Information Specialists, combined with a single tailored mailing of print material; 3) BEM delivered by CIS Information Specialists, combined with four mailings of print material tailored to information obtained solely at baseline; and 4) BEM delivered by CIS Information Specialist, combined with four mailings of print material to information obtained at baseline and at short-term follow-up. Smokers who call the CIS for assistance will be asked questions concerning their state of change, motives for changing, barriers to changing, cessation history, attributions for previous relapses and demographic characteristics. These data will be transferred to the Tailored Message Core facilities, where computer-tailored messages will be created and mailed to the smoker. The impact of the interventions will be assessed three and twelve months after the baseline CIS call using telephone interviews. Results of the trail will provide important information at a crossroad of tailoring research. Nearly all practical formulations of tailored health messages struggle with two questions: "Does sending a series of tailored messages work better than sending a single tailored message?" and "Should we devote the time and effort required collecting new information from the individual for retailoring?" These questions have critical cost implications to administrators of such programs. For this reason, we will conduct a cost-effectiveness analysis of each intervention. It is also for this reason that our sample size requirements allow us to test, with sufficient power, differences between each intervention.